MODELS
AI model catalog
Curated frontier LLMs plus image, video, and audio models from OpenAI, Anthropic, Google, Meta, DeepSeek, Alibaba, xAI, Mistral, and the open-source side — context windows, per-token pricing, modalities, and Chinese-language benchmarks (C-Eval, CMMLU, SuperCLUE) where available.
Claude Opus 4.7
anthropic · claude
Anthropic's top-tier reasoning model for long-context, agentic, and code-heavy work.
Claude Sonnet 4.6
anthropic · claude
Anthropic's mid-tier workhorse for coding, agents, and long-context reasoning.
Claude Haiku 4.5
anthropic · claude
Anthropic's small, fast Claude — cheap tool-calling and vision with 200K context.
Grok 4
xai · grok
xAI's flagship reasoning model with a 256K context and live X integration.
GPT-5
openai · gpt
OpenAI's flagship reasoning model with 400K context and native multimodal I/O.
GPT-5 mini
openai · gpt
A cheaper GPT-5 tuned for high-volume tool use and everyday agent work.
Kimi K2
moonshot · kimi
Open-weight 1T-parameter MoE from Moonshot, tuned for agentic tool use.
Mistral Large 2.5
mistral · mistral
Mistral's flagship text model with strong tool use and European hosting options.
GLM-4.5
zhipu · glm
Open-weight Chinese frontier model with vision and tools at budget pricing.
Gemini 2.5 Pro
google · gemini
A 2M-context multimodal reasoner with native video and audio understanding.
Gemini 2.5 Flash
google · gemini
Cheap, fast multimodal workhorse with a 1M-token window.
Qwen 3
alibaba · qwen
Open-weight multilingual model with vision and tool use at budget pricing.
o3
openai · o-series
OpenAI's heavyweight reasoning model for hard math, code, and multi-step problems.
Llama 4
meta · llama
Meta's open-weight multimodal model with a 1M-token context window.
Mistral Small 3.1
mistral · mistral
Open-weights 24B with vision and a 128K window at bargain prices.
o3-mini
openai · o-series
Cheap reasoning model tuned for STEM, code, and structured tool calls.
DeepSeek R1
deepseek · deepseek
Open-weight reasoning model that matches o1 on math and code at a fraction of the price.
DeepSeek V3
deepseek · deepseek
Open-weight 671B MoE that matches GPT-4o on most tasks at a fraction of the price.
可靈 Kling 1.6
kuaishou · kling
Kuaishou's video model — Chinese-built, cost-leading, viral on TikTok and Douyin.
Veo 2
google-deepmind · veo
Google DeepMind's text-to-video — best-in-class physics and motion realism.
Gemini 2.0 Flash
google · gemini
Cheap, fast multimodal workhorse with a 1M-token context window.
Sora
openai · sora
OpenAI's text-to-video — the brand-name standard, ChatGPT Plus / Pro bundled.
Llama 3.3 70B
meta · llama
Open-weights 70B that matches Llama 3.1 405B quality at a fraction of the cost.
o1
openai · o-series
OpenAI's first reasoning model — the original 'thinks before answering' release.
混元 Hunyuan Large
tencent · hunyuan
Tencent's open-source 389B MoE flagship — biggest open-source MoE at release.
Recraft V3
recraft · recraft
The text-rendering and brand-design specialist — beats Imagen 3 on Artificial Analysis.
Stable Diffusion 3.5 Large
stability · stable-diffusion
Stability's open-weights image model — the workhorse for self-hosting and ComfyUI.
Yi-Lightning
01-ai · yi
Cheap, fast Chinese-first chat model from 01.AI, tuned for high-throughput production use.
Qwen 2.5
alibaba · qwen
Open-weight general-purpose LLM with strong multilingual and coding chops at low cost.
Imagen 3
google · imagen
Google's text-to-image model with strong typography and prompt fidelity.
FLUX.1 Pro
black-forest-labs · flux
Black Forest Labs' top-tier image model — built by Stable Diffusion's original team.
階躍 Step-2
stepfun · step
Stepfun's trillion-parameter Chinese LLM — the dark horse of the PRC race.
訊飛星火 Spark 4.0 Ultra
iflytek · spark
iFlytek's Spark — strongest Chinese ASR/TTS heritage in any LLM.
Claude 3.5 Sonnet
anthropic · claude
The legendary 2024 coding workhorse that defined Claude's developer fanbase.
Runway Gen-3 Alpha
runway · runway-gen
Runway's video model — paired with the best editor UI in the space.
豆包 Doubao Pro
bytedance · doubao
ByteDance's flagship LLM — China's price-war torchbearer.
GPT-4o
openai · gpt
OpenAI's multimodal workhorse handling text, image, and audio in one model.
Llama 3 8B
meta · llama
Meta's mass-deployable 8B open weights — runs anywhere from a laptop up.
Claude 3 Opus
anthropic · claude
Anthropic's first frontier-tier Claude — still on Bedrock for legacy stacks.
Gemini 1.5 Pro
google · gemini
Google's 2024 long-context champion — 2M tokens before anyone else.
text-embedding-3-large
openai · embedding
OpenAI's flagship embedding model for RAG and semantic search.
Mixtral 8x7B
mistral · mistral
Mistral's classic open-source MoE — 47B total, 13B active per token.
Whisper Large v3
openai · whisper
OpenAI's open-source speech-to-text — multilingual, robust, free to self-host.
GPT-4 Turbo
openai · gpt
OpenAI's 2023 flagship — the GPT-4 you remember from before -o.
Yi-34B
01-ai · yi
01.AI's bilingual open-weights model — strong English + Chinese in one stack.
文心 ERNIE 4.0
baidu · ernie
Baidu's flagship LLM, deeply tied to mainland search and enterprise compliance.
DALL-E 3
openai · dall-e
OpenAI's text-to-image — strong prompt fidelity, baked into ChatGPT Plus.
ElevenLabs Multilingual v2
elevenlabs · elevenlabs-tts
ElevenLabs' flagship TTS — most natural English voices, 29 languages.
Falcon 180B
tii · falcon
UAE's TII open-source 180B — the legacy frontier of pre-Llama-3 era.
CLIP ViT-Large
openai · clip
OpenAI's foundational text-image embedding — the bedrock of every diffusion model.